Among individuals with a substance use disorder (SUD), major depressive disorder (MDD) is the most common comorbid Axis I disorder, and is associated with a multitude of negative outcomes, including higher treatment costs, greater risk of suicide attempts, worse overall quality of life, and poorer treatment response. Integrated interventions for concurrent treatment of both disorders have been suggested as the optimal mode of treatment delivery. However, few randomized trials have tested integrated treatment against evidence-based comparison treatments, and integrated interventions are not yet empirically-supported. Treatment researchers have begun focusing on the mediating processes involved in behavioral treatments for SUDs, with hopes of increasing the effectiveness, efficiency, and portability of interventions. By comparison, the existing literature on mediating processes in comorbid populations is strikingly lacking. Discovery of mediating processes should provide insight into the common and unique mechanisms of integrated and non-integrated treatment approaches, informing the development of more powerful interventions. The proposed research will examine mediators of substance use and depression outcomes during a randomized, controlled trial of outpatient group therapies for veterans with comorbid SUD-MDD. Veterans were randomized to either Integrated Cognitive- Behavioral Therapy (ICBT) or Twelve-Step Facilitation (TSF). Key mediators from previous studies of SUDs (self-efficacy, 12-step affiliation, social support for abstinence, coping skills) and MDD (dysfunctional attitudes, negative mood regulation) were assessed quarterly during 6 months of treatment and 12-months of follow-up. Under the first primary aim, latent trajectories of 12-step affiliation, dysfunctional attitudes, and negative mood regulation during treatment will be estimated with latent growth curve models, with the prediction that the TSF group will have increases in 12-step affiliation over time, and change in each mediator will be associated with change in depression. Under the second aim, trajectories of self-efficacy, 12-step affiliation, social support for abstinence, and coping skills during follow-up will be estimated, with the hypothesis that TSF will have reductions in 12-step affiliation, ICBT will maintain levels of self-efficacy, and both groups will have reductions in social support for abstinence and coping skills over time. Changes in each mediator are expected to be associated with substance use trajectories. The third aim will examine the effects of baseline neurocognitive impairment on mediators and their association with outcome. It is predicted that greater neurocognitive impairment will have a deleterious effect on latent trajectories of self-efficacy, dysfunctional attitudes, and negative mood regulation. In addition, the strength of association between these mediators and outcomes will be weaker for patients with greater neurocognitive impairment.